2012 Graduate Workshop in Complexity and Computational Social Science

Student Projects

Each student began a research project during the two-week workshop. Below are brief descriptions of these various projects. These projects will form the basis for dissertation chapters and/or journal articles.

Omar Guerrero, Economics, George Mason University (oguerrer@gmail.com).

Omar is refining our understanding of the dynamics of unemployment and the impact of social ties within, and across, firms in this process. In his model, he makes explicit assumptions about how within-firm social ties create across-firms ties as employees move from one firm to another, based on various foundational economic theories. By linking social ties to employment opportunities, he is able to generate some key patterns that we observe in real systems as well as make a number of novel predictions about key aspects of unemployment.

Varsha Kulkarni, Informatics, Indiana University (vskulkar@indiana.edu).

Varsha is looking at structural changes in social networks. In one model, she considers the relationship between interactions (moderated by social distance) tempered by tolerance and conformity. She finds, using this basic model, that with time the system becomes tightly connected, though simple behavioral changes, like "familiarity breeds contempt," results in a much richer dynamics. In an alternative model, she allows conflict to spread as friends become supportive of one another's battles. Such a world results in the formation of quite separate components, that begin to transform into new components as time begins to heal the original wounds.

Yuntao Long, Economics, Institute of Quantitative and Technical Economics (yuntao.long@gmail.com).

Yuntao is using an agent-based, financial-market model to explore the impact of information flows among investors. In the model, information flows across well-defined networks connecting investors. She finds that asymmetric outcomes are common, and that the network's influence on the flow of information and the subsequent behavior it induces must be addressed in the formation of policy.

Robert Mamada, Sociology, University of Hawaii at Manoa (mamada@hawaii.edu).

Robert is intrigued by the diffusion of innovation. To model such diffusion, he uses partial differential equations based on the heat equation from physics to capture the flow of the various types of adopters in such markets. He introduces a notion of "social temperature'' based on strategy switching, and using this model he tracks the impact of the various types of adopters as they enter the market and encourage further adoption by others as the innovation diffuses across the marketplace.

Felipe Romero, Philosophy, Washington University (cfromero@wustl.edu).

Felipe is interested in understanding how envy influences social systems. In his model, he links notions of envy to agent behavior, and from this he can formulate a dynamic model of agents responding to such incentives as they interact with others in an Ultimatum game. In a dynamic setting, he finds that envy in the system decreases over time (at a decreasing rate), resulting in an overall decline of system-wide inequality. If agents are allowed to form clusters, the system tends to self-organize into local, envy-free zones.

Isaac Sasson, Sociology, University of Texas (isasson@prc.utexas.edu).

Isaac wants to resolve various paradoxes that arise in the statistical analysis of demographic data. For example, while higher social and economic status (SES) tends to improve health outcomes, there are well-documented examples where such trends reverse for certain groups. Models underlying such phenomenon are rife with nonlinear interactions, and hence they tend to be difficult to understand with standard methods. Using new computational tools, he finds that the addition of intra-group heterogeneity introduces a rich set of potential behavior, that may well correspond with the paradoxical patterns that are arising in the data.

Keith Schnakenberg, Political Science, Washington University (keith.schnakenberg@gmail.com).

Kieth considers how agents categorize others and use these categories to mediate strategic interaction. Such models can be applied to a wide variety of phenomena, such as understanding better how issues like race play out in political systems. In the research, agents develop both categories and strategic reactions to this categorization as they interact with one another in a well-defined class of games. Initial results indicate that agents are quite heterogeneous in terms of the number of endogenous categories they form, with the majority of agents relying on only a small number of categories.

Trey Thomas, Government, University of Texas (treythomas@gmail.com).

Trey is exploring social contagion in political agendas. Empirically, various political topics (as measured by, say, the Congressional Record) tend to be characterized by long periods of relatively low interest, punctuated by occasional spikes of intense interest. Using an agent-based model, he finds that he can generate similar patterns of issue interest tied to a small set of critical variables.

Beniamino Volta, Anthropology, University of California (benvolta@gmail.com).

Ben is seeking a generative explanation for the emergence of different types of pre-industrial cities like those found in Mesoamerica. In such systems, we see a variety of well-defined settlement patterns ranging from grid-like cities surrounding a well defined elite core to much less centralized arrangements. To explore such formations, he created a model with two types of agents---elites and commoners---with different incentives for where to locate given the existing pattern of settlement. By varying a few parameters, he is able to generate different settlement patterns that appear to correspond with the key patterns that have been observed in the field.

Alexander van der Vooren, Economics/Sociology, Utrecht University (a.vandervooren@uu.nl).

Alexander wants to understand the dynamics of product offerings in competitive markets with multi-attribute goods and tie these to data he has collected on automobile offerings in the Netherlands. He assumes that new car configurations are constrained by a manufacturer's current offerings, those of its competitors, and the expected demands of consumers. His analysis focuses on the dynamics of the product "clouds" that form over the attribute space of the goods, and he finds that the various product niches are filled at decreasing rates, with manufacturers becoming more specialized as the number of firms increases. He has also been able to generate results about market concentration and shakeouts tied to innovation costs and changes in consumer demands.

John H. Miller , miller@santafe.edu.